Early Agriculturalist Population Diasporas?

Farming, Languages, and Genes

Author(s): Peter Bellwood
Source: Annual Review of Anthropology, Vol. 30 (2001), pp. 181-207
Published by: Annual Reviews
Stable URL: http://www.jstor.org/stable/3069214 .
Accessed: 04/01/2011 00:11

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .
http://www.jstor.org/action/showPublisher?publisherCode=annrevs. .

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Annual Reviews is collaborating with JSTOR to digitize, preserve and extend access to Annual Review of
Anthropology.

http://www.jstor.org
 Annu. Rev. Anthropol.2001. 30:181-207
Copyright( 2001 by AnnualReviews. All rightsreserved

EARLYAGRICULTURALIST
POPULATION
DIASPORAS?FARMING,LANGUAGES,
AND GENES

PeterBellwood
School of Archaeologyand Anthropology,AustralianNational University,CanberraACT
0200, Australia;e-mail:peter.bellwood@anu.edu.au

Key Words spreadof agriculture,Neolithic/Formative archaeology,farmer-hunter

interactions,languagefamilyoriginsanddispersals,populationexpansions
* Abstract Theconsequencesof earlyagricultural developmentin severalregions
of the Old andNew Worldsincludedpopulationgrowth,the spreadof new material
culturesandof food-producing economies,theexpansionsof languagefamilies,andin
many cases the geographicalexpansionsof the earlyfarmingpopulationsthemselves
into territoriespreviouslyoccupiedby huntersand gatherers.This chapterdiscusses
someof thedifferentoutcomesthatcanbe expectedaccordingto thedifferingperspec-
tivesof archaeology,linguistics,andbiologicalanthropology.
I arguethatagriculturalist
expansionlies attherootof manyof theworld'smajorlanguagefamilies,althoughthis
need not implythatfarmersalwaysreplacedhunter-gatherers in the biologicalsense.
History,enviromental variations,andpriorculturalconfigurations dictatedmanyof the
outcomes,some of whichplayeda fundamentalrole in the large-scalegenesis of hu-
manculturalandbiologicalpatterningfromNeolithic/Formative timesinto the world
of today.

INTRODUCTION

This chapter discusses the multidisciplinaryimplications of a series of major

changes from hunting-gatheringto farmingthat occurredin various parts of the
world, at varyingtimes, often independentlyof each other.They are of particular
interestbecause of their potentiallinkages with language family and population
dispersals.They form majorkeystones in the explanationof the patternsof hu-
man variation,both culturaland biological, thatcharacterizedthe pre-Columbian
world and that still dominatein many partsof the world today.As far as the gen-
eral likelihood of early agriculturalist(Neolithic/Formative)populationdispersal
is concerned,the time has come to takestock,to look dispassionatelyat the options,
and to take a worldwidemultidisciplinaryview. Some guidelines and a feeling for
issues thatmight respondto theoreticaldebate shouldhelp the momentumalong.

0084-6570/01/1021-0181$14.00 181
182 BELLWOOD

THE ORIGINSAND DISPERSALSOF AGRICULTURAL

SOCIETIESAND LANGUAGEFAMILIES
Humanprehistorygives us a recordof two very important,yet at first sight unre-
lated,examplesof expansion.These are(a) the expansionsof agriculturalsystems
from hearthareas such as SouthwestAsia, China, and Mesoamerica,and (b) the
expansionsof the world'smajorlanguagefamilies. Some of the latterareof course
associatedwith predominantlyhunter-gatherer populations,butthe majorityoccur
in agriculturallatitudesand theircomponentlanguagesare spokenby people who
were alreadyagriculturalistsat the dawnof history.Manyof these widespreadagri-
culturalistlanguagefamilies, such as Austronesian,Indo-European,Niger-Congo,
Uto-Aztecan, and Afroasiatic,had reachedtheir precolonialgeographicallimits
(give or take a few hundredkilometres) long before the local existence of any
writtenrecords-their spreadsbelong among prehistoricfarmers/pastoralists and
small-scale social formations,ratherthan among the great conquestempires and
charismaticworld religions of history. Could the early dispersals of agriculture
and the early spreadsof certainmajorlanguage families be linked effects of the
same underlyingset of causes? Do these causes relateto the demographicgrowth
and rapidexpansionprofiles of early farmers?
In recent years, a numberof prehistorianshave suggestedthat there are major
linkagesbetweentherelevantcultural,linguistic,andbiologicaldatasetspertaining
to early farming dispersal in differentparts of the world (e.g., Bellwood 1984,
1984-1985, 1991, 1996a,b, 1997b, 2000a; Cavalli-Sforza& Cavalli-Sforza1995;
Diamond 1997a,b, Higham 1996; Renfrew 1987, 1991, 1992, 1994, 1996). The
essential reasoninghere, as clearly set out by Renfrew(e.g., 1996), is thatthe rise
of farming,whereverandwheneverit occurred,formeda veritable"unconformity"
thatran spatiallyacross the chronologicalcourse of history,an unconformitythat
alwayshadthe potentialto forma line of weaknessalong which factorsthatcaused
spreadingcould have operatedwith vigor.
In addition,such spreads,viewed from the experienceof recenthistoricaldias-
poras,would alwayshavehad the potentialfor substantialdegreesof isomorphism
between cultural,linguistic, and biological variation(i.e., if they involved well-
delineatedethnic groups).Furthermore,correlationsbetween these variablesmay
have continuedlong afterdispersalhad ceased. But let it be stressed,given current
touchinessaboutthis issue in the anthropologicalliterature,thatno one seriously
claims absolute isomorphismbetween the data sets (archaeology,language, bi-
ology) in any particularearly farming situation.Language shift, gene flow, and
culturalborrowingcan obviouslyoperatein almostall humancontexts,except per-
haps in situationsof extremeisolation. So the possibility of any 1:1:1 correlation
between a gene pool, a culture, and a language, each changing only by internal
variationof inheritedsource materials,can be dismissed right from the start.
Becauseof suchreal-worldcomplications,the questionsaskedin this essay can-
not be given simple answers.But it is possible to suggest that,in a world peopled
entirelyby huntersand gatherers,any groupswho by some means developed and
 EARLYAGRICULTURALDIASPORAS? 183

came to dependuponsystematicmethodsof food production(of plants,of animals,

and-best of all-of both) would have acquireda "demographicedge" over their
neighbors,particularlyif those neighborsdid not also, by choice or circumstance,
adoptthe food-producingeconomy themselves.Increasingbirthrateswould have
promoteda geographicalbudding-offof the agriculturalpopulationsas theirnum-
bers increased,as long as social or geographiccircumscriptiondid not intervene.
Bearingthis in mind,therewould appearto be two oppositeoutcomesin regions
surroundingearlyfoci of agriculturaldevelopment,with a rangeof possibilitiesin-
between.At one extreme,thehunter-gatherer neighborsof the agriculturalistscould
have adoptedthe agriculturaleconomy themselves,thus inhibitingany tendencies
by the existingfarmersto spreadintonew territory.At the otherextreme,theycould
have remainedas hunter-gatherers facing eventualassimilationinto the largerand
expandingfarmingsocieties. If the formeroptionprevailed,agriculturewouldhave
spreadultimatelythroughmanydiverselinguistic andarchaeologicalpopulations,
previously hunter-gatherers,with similar levels of diversityprobablycontinuing
thereafter.If the latter option prevailed, agriculturewould have spread mainly
throughthe expansionof the farmingpopulationsthemselves, togetherwith their
languages.Undersuchcircumstancesof farmerexpansionfroma homelandregion,
we would expect the impositionof relativehomogeneityof languageandmaterial
cultureon theensuingculturalandlinguisticpatterns,andpossibly a corresponding
spreadof certainpopulation-specificmarkersin genetics.
It is clear, however, that under pre-statesocial conditions, such absolute ex-
tremesarebothmost unlikelyoutcomes.Farmerswill often seek new land;hunters
will often, if allowed, familiarizethemselves with farmingtechniques.Whereex-
actly, within the continuumof reality,differenthistoricaltrajectorieswill fall can
be estimatedfrom several sources of data:
1. Evidence for agricultural"homelands,"and the time and space coordinates
of agriculturalsystem spread into regions beyond those homelands, com-
bined with a considerationof the productive"power"of the various agri-
culturaleconomies and theirresultinglong-termpotentialsfor demographic
expansion.
2. Evidence of pre-agriculturalto agriculturalcontinuity,or hiatus, in the ar-
chaeological record (in general terms, Mesolithic to Neolithic in the Old
World,Archaicto Formativein the New).
3. Patternsof diversity/homogeneitythroughspace in early agriculturalmate-
rial culture.In this regard,some early Neolithic/Formativeculturalassem-
blages are remarkablywidespreadcomparedto preceding and succeeding
patterns,and this is a counter-intuitivesituationif one considersincreasing
sedentismand settlementendogamyto be factorsdifferentiatingearly farm-
ers from theirmore mobile hunter-gatherer predecessors(i.e., early farmers
had good reason to be parochial).
4. Evidence for language family homelands,proto-languageculturalvocabu-
laries (especially agriculturalcognate sets), rates of spreadas derivedfrom
184 BELLWOOD

family tree structures(to be discussed below), and inter-familyborrowing

histories.
5. Palaeoanthropologicalstudies of ancient skeletalremainsfrom the Mesoli-
thic-Neolithictransition,togetherwith genetic evidence for populationhis-
tories in termsof multiplenucleargene clines, mtDNA and Y chromosome
haplogrouphistoriesand theirestimatedcoalescence times.
Thehistoricalandbehavioralpossibilitiesinvolvedin anyhunter-to-farmer tran-
sition, encompassingfactors such as economic challenge and response,language
continuityand languageshift, populationdispersaland gene flow, are so trulyim-
mense on a worldwidescale as to defy brief generalization.This paperaims only
to examine some of the majorarchaeological,linguistic,and genetic datasets and
to applysome linkingtheoryderivedfromethnographicandhistoricalobservation.
Did ancienthunter-gatherers commonlyadoptagriculture,as claimed,for instance,
by Price & Gebauer(1992, Price 1995)? Did ancient (pre-stateand pre-literate)
populationsshift their languagesfrequently?Wereearly farmingpopulationsrel-
atively endogamousor exogamous with respectto neighboringhunter-gatherers?
Did farmersand hunter-gatherersswitch their economies back and forth so fre-
quently that any attempt to differentiatethe two lifestyles becomes pointless?
These are all questions thatrequiretheoreticalconsiderationif interpretationsof
episodes of prehistoricagriculturalexpansionare to carryconviction.

THE PERSPECTIVEFROM ARCHAEOLOGY:

TEMPO, SPREAD, AND FRICTION

Try as they may, archaeologistsalone can never prove that a populationexpan-

sion occurredin prehistorybecausematerialcultureis always capableof diffusing
beyond the hands of its creators.But it is possible to test models created in the
light of datafrom otherdisciplines and to debatehow populationdispersalmight
have been structuredin specific situationsand what kind of archaeologicalevi-
dence could demonstrateits occurrence(Adamset al 1978; Anthony 1990, 1997;
Burmeister2000; Chapman& Hamerow 1997; Rouse 1986). Early agricultural-
ist expansion across continentalspace could have consisted, at one extreme, of
continuouspopulationgrowth along an expansion front, of the type defined by
Ammerman& Cavalli-Sforza(1984) for Neolithic Europeas a wave of advance
fueled by "demic diffusion."At anotherextreme, one might have a progression
of saltatoryjumps from one suitable environmentto another,as suggested for
Neolithic Greece by van Andel & Runnels (1995). These models assume that
early agriculturalpopulationsenteredlandscapeseitherdevoid of humansor with
relatively sparse hunter-gathererpopulations.Were they to enter landscapes al-
ready settledby otherfarmersor by dense and complex hunter-gatherer societies,
then less expansive and more reticulateoutcomes must be expected (Bellwood
1996b).
 DIASPORAS?
EARLYAGRICULTURAL 185

One problem for archaeologistsis that any populationdispersal,by virtue of

imposing a chain of foundereffects and spanninga rangeof environments,as well
as offeringa rangeof outcomesfromincreasedwealthto increaseddeprivation,can
impose changes on the societies concernedsuch that homelandsneed not always
be patentlyobvious in the archaeologicalrecord.Forinstance,populationdispersal
can alter social structuresby encouragingfounder-basedideologies as new bases
for social ranking(Anthony 1990, Bellwood 1996c for Austronesia;K.P. Smith
1995 for Iceland).On the otherhand,Blust (1999) shows how Austronesianpop-
ulations lost many culturalitems as a result of adaptationsduringtheir dispersal
history from Island SoutheastAsia into Polynesia. Populationdispersalcan also
lead to an importantcategory of situationsin which the descendantsof partially
agriculturalsocieties enteradverselandscapesand are therebyobliged to special-
ize "back"into a hunting and gatheringeconomy. Good examples of the latter
include the Punan/Penansago collectors of Borneo, the southernMaoris of New
Zealand,also probablythe Numic- (Uto-Aztecan)speakingpeoples of the Great
Basin (Bellwood 1997a, 2000a, in press; Hill 1999, Sather 1995). Many early
Polynesianagriculturalistpopulationsalso specializedtowarda hunting-gathering
economy among the naive bird faunas of Oceania (Anderson 1989, Steadman
1999), at least until avian extinction/extirpationoccurred and obliged them to
refocus on agriculture.
Such trajectoriesimply that many early Neolithic/Formativecultures did not
initiallycarrytheirfull homelandagriculturaleconomy intonew lands,particularly
when those new landswererelativelymarginalin agriculturalsuitability.Dispersal
is surelya generatorof strongerselectiveandfoundershipfactors,andof morerapid
change,thanis merelystayingat home, a conclusionreachedalso by linguistswith
respectto a greaterrapidityof linguisticchange among small migrantpopulations
(Blust 1991, Fortescue 1997, Ross 1991). The lesson here for archaeology is
that migrantsdo not merely clone themselves and their culturesindefinitely.So
rejectinga dispersal-basedexplanationjust becausea homelandis not immediately
obvious is not alwaysjustified.
But merely consideringthe possible structureof an early farmerdispersal(or
lack thereof)will not proveordisprovethatone occurred.Forthis we need to exam-
ine specifics andto consideraspectsof culturaltransmissionthroughtime. The loci
of primaryagriculturaloriginacrossthe worldarefairlywell establishedwithinthe
archaeologicalcommunity;few would quibbleover the significanceof the Fertile
Crescent,centralChina(middle Yangzi and Yellow basins), highlandMesoamer-
ica and the PeruvianAndes, or perhapseven the EasternWoodlandsof the USA
(Harris 1996, Price & Gebauer 1995, B. Smith 1995). Less agreement would
be forthcomingover the roles of West Africa and the Sahel/savanna,Amazonia,
and the New GuineaHighlands.The issue here is not over the homelandsof spe-
cific crops, since many crops originatedin regions not known to have harbored
independenttransitionsto agriculture(e.g., India,SoutheastAsia). The issue con-
cernsindependenceof the trajectoryfromforagingto farming;archaeologicaland
botanicaldemonstrationof this can be claimed only for SouthwestAsia, with the
186 BELLWOOD

otherregions (includingeven China)still in the realmof extremelikelihoodrather

thancertainty.
A numberof questionsnow need to be asked,from a comparativeperspective.
First, how productive(food outputper head of population)were the agricultural
systemsthatevolved in the variousregions?This is an importantmatterto consider
for any demographicgrowthmodeling.Therecan be no doubtthatthe remarkably
multifacetedagropastoraleconomy thathad evolved in SouthwestAsia by the end
of the Pre-PotteryNeolithic phase (c.7500 B.C.E.) wins in terms of sheer number
of majordomesticatedcereals and animals.China,with its rice, bovids, pigs, and
poultrycomes a close second. CentralAfrica, with its millets and the adoptionof
caprinesand bovids from the north,would probablycome third.New Guineaand
the Americas,withno significantmeatanimalsat all (pigs arepost-3000 BP in New
Guinea)anda relativelygreateremphasison tubersas opposedto cereals,all come
furtherdownthe scale. The fact thatthe complex civilizationsandfarmingcultures
of the Americasdependedto a high degree on only one majorcereal (maize) and
essentially on huntedmeat (localized stocks of camelids, dogs, guinea pigs, and
turkeysnotwithstanding)could have had a dramaticimpact on many aspects of
cultureandhistory(e.g., Harris1978, pp. 99-110). In termsof the overallextentsof
languagefamily and agriculturalsystem expansionfrom these homelandregions,
I suspect a similarrankingapplies. The evidence for early farmingdispersalout
of SouthwestAsia and Chinais far more compelling across all disciplines thanit
is out of the New GuineaHighlandsand the Andes.
A second question arises immediately,and this concerns tempos of spread.
How long did it take for the agriculturalcomplexes that developed in the various
homeland regions to reach their prehistoriclimits? I am a firm believer in the
concept of a "Neolithic Revolution,"insofar as it relates to the inception and
economic dominationof cereal cultivationin Mesoamerica,SouthwestAsia, and
China(Sherratt1997). In specificcircumstances,the conceptremainsvery apt.But
tempos of origin and subsequenttrajectoriesof spreadmust be kept analytically
separate.We have the following well-establishedspreadtimes from sourceregion
to ultimateprehistoricgeographicallimitdemonstratedby archaeologyandcareful
use of chronologicaltechniques,well-establishedregardlessof who (expanding
farmersor convertinghunters)actuallyspreadthe farmingway of life:

* Fertile Crescentto Britain, 3000 years (7000 to 4000 B.C.E.)over 3600 km

(Price 2000);
* YangziBasin to IslandSoutheastAsia, 4000 years(6500 to 2500 B.C.E.)over
5000 km (Bellwood 1997a);
* CentralMesoamericato the Southwest,2000 years(3500 to 1500 B.C.E.)over
2500 km (Muro 1998, Benz & Long 2000);
* PakistanintoPeninsularIndia,4000 years(7000 to 3000 B.C.E.)over2000 km
(Chakrabarti1999).

On these scales, average rates of spreadrangedbetween 0.5 and 1.25 km per

year, which can surely be consideredfairly slow. It is irrelevanthere whetherthe
 EARLYAGRICULTURAL
DIASPORAS? 187

agriculturewas being spreadby convertinghunter-gatherersor range-expanding

farmers-both groupswould have become subjectto populationincreasein good
environments.Recent colonizing populationsin low-density situations of good
health and rich resources were frequently capable of doubling numbers every
generation,andthereis no reasonwhy earlyfarmers,whatevertheirorigins,should
have been differenton the large scale. Something seems to have been applyinga
braketo the operationof free and untrammeledfecundity.
That something was presumably,on the continentsat least, a combinationof
environmentalvariationandnativehunter-gatherer resistance.The farmingway of
life clearly had to progressthroughsome very complex environmentaland social
barrierzones. Environmentally,these included alterationsof rainfallseasonality,
as for instancein the changes from winterto summer(monsoon) rainfallin India
and Sub-SaharanAfrica, and the latitudinalfactorsof day-lengthand temperature
variation,significant in the East African, East Asian, and American situations.
Crosby(1986, p. 18) andDiamond(1997a, p. 176) have suggestedthatlongitudinal
(north-south)geographicalaxes led to slowerspreadthanlatitudinalones, butwhile
this may be generallytrue it does not work in all instances (compareSouth Asia
with easternand southernAfrica). The conclusion at this point must be that, over
the long term and the long distance, agriculturedid not always spreadquickly,
whateverthe axis.
However,the situationlooks ratherdifferentif we focus on more specific sit-
uations (Figure 1). Intermediate-scaleinstances of rapid spread of both initial
agricultureand the archaeologicalassemblagesattachedto it are in fact relatively
common,andit is herethataxes can matterbecauseall suchrapidspreadsoccurred
in zones of ratherhigh suitabilityfor the agriculturalsystems concerned:

* BismarckArchipelagoto westernPolynesia, 500 years (1300 to 800 B.C.E.)

over a fairly daunting4500 km, but mostly over water and along the same
tropicallatitude(Lapitaarchaeologicalcomplex-Kirch 1997);
* HungarianPlain to Alsace, 400 years (5700 to 5300 B.C.E.)over 1000 km,
againalong the same generallatitudeandthroughthe same zone of temperate
climate (LBK [Danubian]archaeologicalcomplex-Keeley 1992, Bogucki
1996, Gronenbor 1999);
* Levantto northwesternPakistan,500 years or less (7500 to 7000 B.C.E.)over
2500 km, again along one latitude and in one winter rainfallclimate zone
(late Pre-PotteryNeolithic archaeologicalcomplex-Bar-Yosef 1998);
* East African Lakes to South Africa, 1000 years (1000 B.C.E.to C.E.1) over
3500 km through one zone of predominantlymonsoonal rainfall, albeit a
north-southone in this instance (Chifumbazecomplex-Phillipson 1985,
1993, Ehret 1998).

Averagerates of spreadin these cases were much fasterthan in the first more
generalized group, ranging from 2.5 to 5 km per year in the continentalcases,
with the fastest (9 km per year) predictablyin Oceania. In general,most spreads
werepredominantlylatitudinal,except for thatin Africa.Clearly,earlyagricultural
Figure 1 The distributionof prehistoricagriculture,with some widespreadprehistoricarcha
associatedwithearlyagriculturalist
expansion.
 EARLYAGRICULTURAL
DIASPORAS? 189

economies and their associated materialculturescould spreadextremelyrapidly

in those environmentsin which they had developedand/orto which they were well
adapted.But environmentalboundariesand transitionsslowed down the process
by creatingfrictionagainstthe easy and unopposedspreadof early farmingcom-
munities, and one may surmisethat such "frictionzones" served as likely arenas
of majorreticulationand reformulationwithin the linguistic and genetic realms,
as well as in material (archaeological)culture. These friction zones, of course,
occurredaroundthe edges of the "spreadzones" (afterNichols 1992) just listed,
in regions such as northernand westernEurope,lowlandNew Guinea,the Indus-
to-Ganges transitionfrom winter to summerrainfallclimate, and the nonalluvial
hilly terrainsof southernChina.We meet themalso in the Americas-for example,
the Amazonianinterfluvesand the easternGreatPlains. We can expect zones of
friction to occur not just in environmentallyunsuitablesituationssuch as semi-
deserts,drygrasslands,or coniferousforests,butalso wherehunter-gatherers lived
in high densities, along productivecoastlines for instance.
The above reasoning suggests that zones of rapid agriculturalspread should
reveal some degree of phylogenetic interruptionto the continuous inheritance
of materialculture across the transitionto agriculture.This should be the case
regardlessof whetherexisting huntersadoptedfarmingrapidly and then spread
geographically,or whetherexisting farmersmigratedin. Rosenberg,for instance,
actually lists hunter-gatherer-to-farmertransitionperiods in many parts of the
world as the majorityof his cases of rapidand discontinuousbouts of punctuated
culturalevolution,when periodsof "stress-generatedsystemicfailure"(Rosenberg
1994, p. 322) stimulatedthe creationof new culturalformations.Continuityover
the whole of a spreadzone from Mesolithic/Archaicforebearswould be a little
unlikely from this perspectivebecause spreadingand continuityobviously con-
tradicteach other in their implications.On the other hand, friction zones should
witness slower trajectoriesfrom hunterto farmer,with many apparentsituations
of regionalcontinuityin cultureand biology. It is here we might expect to see the
clearestevidence of regionalhunter-gatherer adoptionof agriculturevia the phases
termedavailability,substitution,andconsolidationby Zvelebil & Rowley-Conwy
(1986, Zvelebil 1998).
On the ground,of course,it is sometimesvery difficultto ascertainif the change
from Mesolithic to Neolithic or Archaic to Formativein a particularregion does
or does not reveal a phylogenetic break.Opinions on such mattersdiffer greatly
and often become quite heated, as in the debate over the Neolithic transitionin
Europe,where worldsoften collide ratherroughly.The extentof earliestNeolithic
cultures in any situation clearly matters-if they are extremely widespread, as
are the LBK and Lapita, then population movement will always deserve care-
ful consideration.In my own research area, Island Southeast Asia, I have lit-
tle doubt that the widespreadand polythetic Neolithic culturalcomplex with its
rice, pig, and dog bones, red-slippedor paddle-impressedpottery,polished stone
adzes, shell and stone ornaments,barkclothbeaters, and occasional microblade
industriesand spindle whorls, marksa fundamentalphylogeneticbreak,in both a
190 BELLWOOD

material-cultureandaneconomic sense, withtheprecedinglate-Preceramicassem-

blages (Bellwood 1997a). The same applies to the contemporaryLapitacomplex
in Melanesia(Spriggs 1996). In manyotherpartsof the archaeologicalworldsuch
breakshave been claimed and counterclaimedwith such an intensitythat any at-
tempthere to summarizewould be fruitless.In terms of continuity,however,one
importantpitfall must be contemplated.
This is the pitfall of ambiguousphylogeny-the situationin which there ap-
pears to be regionalcontinuitywithin an archaeologicalrecord,but in which the
real homelandis actuallyanotherregion,perhapsone quite distant,with a similar
antecedentmaterialculture.Ambiguousphylogeny becomes a problemin situa-
tionswherethe materialcultureof the immediatelyprefarmingphaseis complacent
over very largeregions in termsof stylistic variation,as it is for instancein much
of SoutheastAsia and the westernPacific (Spriggs 1996). In much of this region,
small flake industriesoften continueapparentlywithouta breakfrom Preceramic
into Neolithic assemblages,but since such flaketools are virtuallyuniversalprior
to the common use of iron, this need mean little. Such a circumstancecan only be
takento imply continuityif it is supportedby otherdata sets.
A finalconsideration,while dealingwith archaeologicaldata,is thatof ultimate
causation of agriculturalspread.Farmerpopulationgrowth and hunter-gatherer
adoptiondoubtless fueled the process to varyingdegrees, but one importantfac-
tor may also be that of the declining homeland environmentsof early farmers.
Some regions of early agriculture,in particularthe Levant(Rollefson & Kohler-
Rollefson 1993) and northernGreece (van Andel et al 1990), are known to have
sufferedenvironmentaldamage duringthe Neolithic, and one wonders whether
this could have been a relevantfactorin inducing spreadsof agriculturalpopula-
tions. The otherregions of early agricultureare less well understoodin terms of
environmentalhistory,but southerncoastal China(particularlyFujian),the Sahel,
northernMesoamericaand drier regions of the Andes could have been affected
very early on by intensive cultivation,just as were Sumer,the Indus Valley, the
SouthwesternUnited States, and EasterIslandin laterculturalphases. Is it purely
coincidentalthat the spreadof Neolithic agricultureinto the Nile Valley and into
Europe (beyond Cyprus and Greece) occurrednot with the first farming in the
Levant(c. 8000 B.C.E.),but with fully agropastoraland pottery-usingpopulations
at about 6500 B.C.E. or later?

THE PERSPECTIVESFROM HISTORY AND

ANTHROPOLOGY: ADOPTING AGRICULTURE

The historicalrecordobviously tells us a greatdeal aboutthe dispersalof farmers

into new lands. It is not my intentionhere to argue whetherthe recent colonial
occupationsof the Americasand Australasiaoffer relevantcomparisonsfor deep
Neolithic prehistory,but there are two observationsI think are of great impor-
tance. The first, most clearly stated by Crosby (1986), is that colonial-era mi-
grantsrapidlydominatedwherepre-existingpopulationswere small,but migrants
 EARLYAGRICULTURAL
DIASPORAS? 191

becameincreasinglyless successfulas priorpopulationsbecamemorecomplexand

numerousand as the territorialgoals became more tropicaland disease-protected.
Thus, Spanishancestrydominatesthe genes and culturesof modem Argentinato
a far greaterextent thanthose of ruralMexico. Britishancestrystill dominatesthe
genes and cultures(not to mentionthe surnamesand place-names)of Australiato
a far greaterextent than those of Indiaor Malaysia.The conclusion here must be
that colonizing populationsof farmerswill always face relativelyresistantnative
cultures,and in the resultingmix many factors will matter:local patternsof dis-
ease, relative demographicprofiles, overlaps in territorialrequirements,relative
complexities of social organization,etc. In the case of Neolithic farmersentering
hunter-gathererterritories,the situationmight seem simple, until one begins to
thinkaboutthe complexityof linguistic,genetic, and social relationshipsbetween
Aboriginesand Europeansin many partsof moder Australia.
The second observationfrom the diasporasof recent history is one that goes
againstthe feeling held by many anthropologiststhatrace, language, and culture
never correlate and always vary independentlyof each other ["It is customary
to insist on the mutual independenceof racial, cultural and linguistic factors"
(Sapir 1918:10)]. For much of humanprehistory,and in certainspecific and well-
knownethnographicsituationssuchas lowlandMelanesia,manyregionsof Africa,
Madagascar,and the Caribbean,a substantiallack of isomorphismcertainlyholds
good. Butin manycolonial-eracases of rapidlong-distancemigration,it manifestly
did not. Naturally,intermarriagewith priorinhabitantsof a region will always oc-
cur, but the fact remains that, in inception, any successful colonizing expansion
will have a core populationwithin which biological variation,language, and ma-
terial culturewill sit togetherisomorphicallyto a very comfortabledegree. Any
such isomorphism,of course, may not last for long, but its long-termsignificance
can still be very greatindeed.
The anthropological(ethnographic)record also provides food for thought
on threefurtherissues. Admittedly,ethnographydoes not revealany situationspar-
allel to the initial spreadsof farmersor languagefamilies (Fix 1999:150). It covers
far too shorta time span and deals for the most partwith societies in retreatfrom
colonial confrontations.But observations(expressedin the ethnographicpresent)
from a fairly broadcross-sectionof the literatureindicatethat:

* Hunter-gatherer and agriculturaleconomies rarelyblend on a balancedbasis

overthe long term(Murdock1967, Hunn& Williams 1982). Manytraditional
hunter-gatherers certainlyhusbandand protectresources.Most farmershunt
andgatherif theirenvironmentsallow such luxuries(Kent 1989), andclearly
all farmersin the Americascontinuedthroughoutprehistoryto be unusually
dependentupon hunting. But 50:50 balances are rare, and those that occur
generallycarrya marginalairfromthe perspectivesof agriculturalexpansion
and demographicgrowth.
* Hunter-gatherersrarelyadoptagriculturesuccessfully.When they come into
contact with farmersand are able to avoid assimilation,they often develop
mutualistictrade-and labor-basednetworksof interaction(Peterson 1978).
192 BELLWOOD

One can of course suggest that modem hunter-gatherers,by definitionand

becauseof encapsulation(Woodbur 1982), only exist becausethey have not
adopted agriculture;hence they are irrelevantfor any considerationof the
behaviorof ancienthunter-gatherers, especially those in environmentswith
high agriculturalpotentials.This observation,however,collides ratherhard
with one majorobjection.
The objectionis thatmany recenthunter-gatherers in richpotential farmlands
were in contactwith farmers,were not encapsulated,yet nevershowedthe slightest
interestin adoptingagriculture.These include the peoples of Californiaand the
NorthwestCoast,andof coursemuchof northernAustralia.One can bringin many
opinions here as to why agriculturewas never adopted-desires for mobility and
nonaccumulationof debts and property,lack of economic need (hunter-gatherer
"affluence")and so forth-but facts remainfacts. Neither northernAustralianor
Californiawerehostileenvironmentsfor agriculture.These populationswereprob-
ably behaving quite logically-why take on the schedulingdemandsof agricul-
tureunless it was necessary?Such examples, of course, make us wonderjust how
frequentlyhunter-gathererswould have adopted agriculturein the deeper past,
especially in the zones of rapidfarmingspread.
Because I have a more detailed survey of all these topics in preparation,I
concludethis section simplyby statingthat,in the ethnographicrecord,huntersand
gatherershave rarelyadoptedagriculture,andif they were to do so, particularlyin
richenvironmentssuitablefor farmerspread,thenquickreactionswouldhavebeen
required.Mesolithic huntersin Baltic Europedoubtless had far longer to weigh
up theiroptionsthandid theircontemporariesin the Danubianloesslands (Keeley
1992, Zvelebil 1998). Huntersin the friction zones marginalfor agropastoralism
might have occupied ethnographicallyunusualoscillating economic trajectories
between farmingand food production(see Kent 1992), but I certainlydoubt that
such situationscould ever have been common. Any assumptionthatall huntersin
prehistorywould or could simply have adoptedagricultureas soon as the word
was whisperedin theirears is probablymisguided.

THE PERSPECTIVEFROM LINGUISTICS: NATIVE

SPEAKERSVERSUS LANGUAGE SHIFT

Linguists debate issues such as language family homeland options and proto-
language dispersal histories, but they rarely focus on the social conditions in
which the dispersals might have occurred.In this regardthey are the opposite
of archaeologists,who place great stress on the ancient societies but often have
unrealisticviews about how languages are transmittedthroughspace and time.
As Nettle (1996, 1999) points out, very little literatureexists on the formationof
language groups in anthropology.For instance, many archaeologistsfavor con-
vergence, creolization, lingua francas, and multilingualismas environmentsfor
 DIASPORAS?
EARLYAGRICULTURAL 193

the spreads of linguistic entities such as Proto-Indo-European(Zvelebil 1995),

Proto-Bantu(Hall 1990), and Proto-Austronesian(Meacham 1984-1985). Such
concepts, often linked with substantialepisodes of language shift, allow archae-
ologists to escape the stigma attachedto migration-basedexplanations.Indeed,
language shift explanationsfor large-scale spreadhave also been proposedby a
numberof linguists (e.g., Nichols 1997, 1998 for Indo-European).
However,in the light of comparativehistoricaland ethnographicobservations,
it seems that creolization, lingua francas, and multilingualismalone cannot be
sufficient explanationsfor such enormous vernacularspreads as those required
to understandthe genetic foundationsof, for instance, the early Indo-European,
Bantu,andAustronesianlanguages(e.g., Mallory1987,pp. 258-59, Vansina1990,
Ross 1997b).These families arenot reputed,on an overallbasis, to revealmassive
tracesof substratumresidueor universalhistoriesof creolization.The latter,apart
from being in large part a postcolonial phenomenonattachedto forced translo-
cation, cannot account for the spreadof language families that can be classified
genetically(Muhlhausler1986, Thomason& Kaufman1988). This need not imply
that individualproto-languagescan never have been pidgins or creoles, although
my understandingof human history suggests to me that translocativesituations
conducive to pidgin formationwould be most unusualin a Neolithic/Formative
context. Pidginization,of course, must be distinguishedfrom the more normal
forms of contact-inducedchange discussed by many linguists (e.g., Thomason&
Kaufman1988, Dutton 1995, Ross 1996, 1997a).
Likewise, lingua francasand the languagesof small rulingelites have not ever,
in history,led to the spreadsof single languageson anythinglike the requiredscale
withoutsubstantialcomponentsof native speakermigrationand settlement,as we
know from the spreadsof Latin,Hellenistic Greek,Spanish,Arabic(Pentz 1992),
andEnglish.The languagesassociatedwith noncolonizingconquestempires,such
as Mongolian,Persian,andNahuatl,did not replaceall the vernacularlanguagesin
the regionsconquered,andsurelythe reasonsfor the greatersuccess of Quechuain
this regardrelateto the Inca policy of populationtranslocationandthe subsequent
adoptionof Quechuaas a linguafrancaby Spanishmissionaries(Heath& Laprade
1982). The spreadsof politically neutralnationallanguages such as Bahasa In-
donesia (Errington1998) and Tok Pisin (Kulick 1992) representcircumstances
specific to moder literatenationswith centralgovernmentsandpositive language
policies, hardlythe assumedcharacteristicsof Neolithic/Formativenonliterateand
pre-statesocieties.
Multilingualism,finally,has traditionallybeen a vehicle for languagemainte-
nance, not replacement,as is clear in linguistically complex ethnographicsitua-
tions in western Melanesia (Kulick 1992) and the Vaupesregion of Amazonia.
In the latter case, an almost unprecedentedrate of linguistic exogamy does not
cause large-scale language mixing or replacement(Sorenson 1982, Aikhenvald
1996). Multilingualismcan lubricatethe workingsof contact-inducedchange and
language shift, but it does not in itself promote the latter and can often militate
194 BELLWOOD

againstit (Sauder 1990). Languageloyalty is also an extremelyimportantfactor

because it can serve as an antidoteto free-wheeling language shift (for interest-
ing case studies, see Schooling 1990, Smalley 1994). Is it realisticto assume that
languageshift alone could have swept throughvast areasof pre-state"Neolithic"
social landscape,in the completeabsenceof any state-levelmechanismsinvolving
literacy,political unity,and the repressionof ethnic identity?
Fora morerealisticreconstructionof languagefamilydispersalhistory,it is nec-
essary to recognize that languagefamilies are genetic entities, transmittedessen-
tially andcontinuouslythroughsuccessive generationsof nativespeakers.Overall,
they aredivergenceratherthanconvergencephenomena;unrelatedlanguagescan-
not convergeinto a geneticallyconstitutedfamily with a commonproto-language.
Presumably,they have originatedin processes of language spread from home-
land regions, ramifyingthroughtime via nodes (proto-languages)delineatedby
groupingsof sharedinnovations(Blust 1995a,b,Peiros 1998).
However, languages rarely split irrevocablyto form innovation-definedsub-
groupsunless some long-distancemovementor intermediatelanguageextinction
occurs. Following initial expansion, chains and meshes will form, along which
innovationsdevelop in overlappingsets (see Pawley & Ross 1995 on innovation-
defined versus innovation-linkedsubgroups).Such chains can in time generate
discrete subgroups,but if the initial spreadis rapidand geographicallyextensive,
then any such subgroupswill show a ratherfrustratingrake-likephylogeny,which
can be uninformativewith respect to homelandand directionof dispersal (Ross
1997a, Pawley 1999). Language families with clearly bifurcatinghistories and
obvious homelands are rare in reality, a circumstancewhich might suggest that
most hadfairlypunctuated(rake-like)ratherthangradualist(tree-like)origins and
early dispersalhistories,as perceivedby Dixon (1997).
To explain such punctuationeffects, it is necessary to resort to social expla-
nations and of course to examine the archaeologicalrecord.Can the concepts of
spreadand frictionzones be appliedto languagehistory,and if so, do the zones so
defined correspondto any degree with those recognized from the archaeological
record?It is essential to rememberhere thatthe comparativelinguistic recordfor
most parts of the world draws on living languages;thus historical leveling and
replacementscan alterthe picturesubstantially(e.g., the expansionsof Turkishin
Anatolia,Sinitic languagesin southernChina,Arabic,English, Spanish,etc.), just
as transformation processescan alterthe archaeologicalrecord.Buteven withthese
provisos, it is apparentthatmanyregions with very strongindicationsof contact-
inducedlinguisticchangeor substratuminterferencetendto correlatewith regions
identifiedas friction zones in the archaeologicaldiscussion above. Such regions
include northernIndia and peripheralregions of Europe within Indo-European
(Masica 1978, Sverdrup& Guardans1999, Wiik 2000), westernMelanesiawithin
Austronesian(Ross 1988, Pawley & Ross 1993, 1995, Dutton 1995), and southern
Chinawithin Sino-Tibetan(Ballard1981).
Conversely,in many archaeologicalspreadzones, the dispersalsof majorlan-
guage families have left few clear traces of any prior linguistic patterning-we
 EARLYAGRICULTURAL
DIASPORAS? 195

often seem to have a series of clean sweeps with no survival of linguistic iso-
lates or major traces of substrata.Such is certainly the case with Austronesian
in most of Island SoutheastAsia, with the Bantu languages in Africa, and with
Sinitic in central China. Of course, linguists have also toyed with the concepts
of spreadand friction (otherwise residual) zones, in particularJohannaNichols
(1992, 1997), with whom the concepts appearto have originated.Like Nichols, I
regardspreadzones as canvases for rapid and relativelyoverwhelminglanguage
movement and replacement(as we would expect from early farming dispersal,
among other reasons), whereas the residualzones of Nichols 1992 can have two
distinct types of origin. They can be end-of-the-lineregions of inflow and sub-
stratumresidue, as in the concept of the friction zone presentedabove. This is
the sense in which Nichols generallyuses the term "residualzone." On the other
hand, many regions of great diversity at the level of whole language families-
areas such as the Middle East, Mesoamerica,East Asia in general, and central
Africa-cannot really be consideredresidualzones but ratherare "upwelling"or
"starburst"zones of net populationincrease and outflow. These regions are all
agriculturalhomelands,andall have linguisticprofilesthatreflectlanguagefamily
genesis and outflowratherthanresidualaccretion(Bellwood 1991, 1994, 1996a,
1997b).
So, in terms of language dispersal,we have three concepts: (a) the homeland
starburstzone of language outflow and nonreplacement;(b) the spread zone of
rapid language flow and widespreadreplacement;and (c) the friction zone of
reticulation.Throughthese zones, the ancestralgenetic componentsof the major
language families must have been transportedfor the most part in the mouths
of native speakers,and processes involving language shift would have operated
most frequentlyin the friction zones. But even in spreadzones, societies would
havebeenpermeablewithrespectto the incorporationof outsiders,perhapsin large
numbersin situationsof low populationdensity,with lack of conflictover landand
bilateralas opposedto tightlyunilineallandownership.Fromthisperspective,early
language dispersals such as those of early Indo-Europeanor early Austronesian
must have involved the movement of sizeable populations of native speakers,
howeverthe criterionof "nativeness"mightbe spelt out in reality.As Ross (1997b,
p. 183) points out for Austronesian:
... it is indeed difficult to conceive of the movement of Austronesianlan-
guages only or even largely in termsof language shift: what could have mo-
tivatedgroupaftergroupto abandonits languagein favorof an Austronesian
one? We must infer thatmovementsof people have played a large role in the
dispersalof the Austronesianlanguages.
This backgrounddebatebringsus aroundagainto the centralhypothesis-that
the foundation dispersalsof the majoragriculturalistlanguage families (i.e., the
spreadof theirbasal-nodeproto-languages)have a high chance of being directly
associatedwiththe spreadof initialfarmingpopulationsthroughregionspreviously
occupiedby hunter-gatherers.If thishypothesishas a fairchanceof survivingcloser
196 BELLWOOD

examination(it goes without saying that absoluteproof will never be an option),

then we would expect a numberof correlationsto occur:
* Language families with potential for this kind of origin should have
strongproto-languagecognate sets, with stable meanings,relatingto crops,
agriculturalactivities, and perhapsdomestic animals, accordingto regional
circumstances.
* They should have indicationsof an early spreadover a large area,probably
to be indicatedby a rake-likebasal phylogeny (i.e., two or more first-order
subgroupsthatcannotbe orderedhierarchicallyandthatcould have emerged
from a priorinnovation-linkedcontinuum).
* They should have time-depthsthat correspondroughly with those for early
agriculturaldispersal from the archaeologicalrecord (issues of language-
family time depthscannotbe examinedin detail here, but see Renfrewet al
2000 for a constellationof currentviews).
* They should have other materialitems reconstructiblewithin their proto-
language etyma that can be correlatedwith nonuniversalitems in the ar-
chaeological record (for many examples of this kind of reconstructionin
the Austronesianfamily, see Pawley & Ross 1994, Ross et al 1998, Kirch
& Green 2001-horizon-like and suddenappearancesare perhapsthe most
useful). The archaeologicalrecordshouldrevealhomelandanddispersalhis-
toriesfor agriculture,andmost importantly,it shouldplace thehomelandsand
directionsof spreadin a geographicalframeworksimilarto the (potentially
linked) linguistic homelandsand spreads.
A numberof pointsrequirespecialcommenthere.First,it is extremelydifficult
to offer homelandsfor languagefamilies thatbeganwith far-flungproto-language
dispersals(Fix 1999, p. 163, makes the same point for biological anthropology-
thatrapidpopulationexpansiontends to mask informationon phylogeny). But in
previous papers,Renfrew and I (see listed references)have suggested that agri-
culturalhomelandsdo correlateto a majordegree with potentiallanguagefamily
homelandsandthatthe dispersalof these languagefamiliescan be seen as radiative
or flower-like,spreadingout of a sourceregion(see also Sherratt& Sherratt1988).
Thereis no doubtscope for disagreementhere,particularlyoverhomelandsof very
widespreadfamilies such as Indo-Europeanor Uto-Aztecan,but I detect increas-
ing agreementamong linguists and archaeologiststhat, for instance, Proto-Indo-
Europeanoriginatedin the vicinity of Anatolia (Dolgopolsky 1993, Gamkrelidze
& Ivanov 1995, Mallory 1997, Renfrew 1999) as opposed to the Ukraine, and
that Proto-Uto-Aztecanoriginatedin Mesoamerica(Bellwood 2000a, Hill 1999)
as opposedto Oregonor California.Both AnatoliaandMesoamericaareof course
well-establishedfoci of early agriculturaldevelopment.The Ukraineand Oregon
are not.
Finally, we must ask if it is purely coincidentalthat most language families
would appearto belong to the past 8,000 years, not to the past 15,000, at least
 EARLYAGRICULTURAL
DIASPORAS? 197

accordingto a very broadrangeof calculationsfrom glottochronologyto rule-of-

thumbcomparisonsof modem languageswith theirancientliteraryforebears.The
only language family commonly given a time-depthof more than 8,000 years is
Afroasiatic.The vast majorityof othersarereputedto be between2,000 and 8,000
yearsold (Bellwood 2000a). Why?Is it becausecomparativelinguisticshas a rock-
bottomlimit of 8,000 years?Or is it because these languagefamilies do represent
real flowerings imposed over a prior continuumof Palaeolithic/Mesolithiclan-
guages, which were not orderedinto sharplyboundedfamilies andwhich, instead,
reflected a vast mesh of intersectingrelationshipssimilar to the Pama-Nyungan
hunter-gathererlanguages of Australia? [I make this point without wishing to
be drawninto the debate over Pama-Nyunganidentity and origins (Dixon 1997,
McConvell1996, Evans& Jones 1997)]. If suchfloweringsdid occurout of agricul-
turalhomelandregions,thenthis could providethe historicalbackgroundfor those
shadowy macrofamilies,such as Nostraticand Austric, which currentlygenerate
so much heat among linguists (e.g., Renfrew& Nettle 1999).

THE PERSPECTIVE
FROMBIOLOGICAL
ANTHROPOLOGY:
BONES,GENES,AND CHRONOLOGY

Early farmingexpansion implies dispersalsof real populations.The only way to

recoverprehistoricpeopledirectlyis throughtheirbones-languages, archaeology,
and the genetics of living populationscan never offer more than proxy data for
the on-the-spotprehistoricculture-wieldinganimalitself. Did the firstfarmersall
radiateas distinct ethnic and racial groups from agriculturalhomelands?Did no
one move at all, with only languagesand agriculturepassing fromgroupto group?
Orwas therealways a mixture,with farmerdemic spreadbeing most markedin the
spreadzones, but hunter-gatherer incorporationbeing most markedin the friction
zones where farmingwas not always the ultimatepanacea?
Palaeoanthropologyshould, in theory,be a majorarbiter.However,it presents
some majorproblems.Large skeletal assemblages are widely reportedfrom Ne-
olithic sites but are rarefrom Mesolithic ones, and when they are availablefrom
either side of the transition,they are usually not from the same site or immedi-
ate region. One interestingexception here is the large Hoabinhianand Neolithic
sample from the shelter of Gua Cha in Malaysia (Bulbeck 2000), where, as one
might expect, the evidence for population continuity is quite strong (the inte-
rior Malay Peninsularrainforestsare undoubtedlya majorfriction zone with re-
spect to agriculture).In Europe,Mesolithic samples are rareand small, and opin-
ions on populationcontinuityor lack thereof vary. For instance, Fox (1996) and
Jackeset al (1997) offer opposingopinionsfor Portugal,a presumedfrictionzone,
whereas Vend (1988) favors Neolithic replacementin the spreadzone of central
Europe.For China, Brown (1998) points to major morphologicalchange across
the Mesolithic/Neolithic boundary.A large literatureon palaeoanthropological
aspects of the Jomon to Yayoi transitionin Japan,recently reviewed by Hudson
198 BELLWOOD

(1999), indicatesmajormorphologicalchange associatedwith Yayoi immigration

from Korea.To my knowledge, however,there have been very few large-sample
studies of morphologicalvariationin human bones dating to either side of the
transitionfrom single localities. Indeed, there is a majorneed for a synthesis of
the palaeoanthropologicalrecord for the hunter-to-farmertransitionon a world-
wide basis.
Three other aspects of palaeoanthropologyare also of direct relevance in the
transitionto farming:trends in health, fertility, and diet. With regardto health,
there is no strongevidence to suggest that early farmingwas the universalslide
into ill-healthsometimesvisualizedas an epiphenomenonof the "affluentforager"
syndrome, at least not until the emergence of crowd diseases and burgeoning
populationdensities. However,the proviso here must be that large samples from
relevanttime periodsarerare.Thatfarmerhealthdeclined in generalthroughlater
prehistoryis not in debate (Larsen 1995) and is not here considered a relevant
issue.
Majorproblemswith determiningthe fertilityprofilesof cemeterypopulations
(Wood et al 1992, Meindl & Russell 1998, p. 390) rendercomment on issues of
earlyfarmerdemographyandbirthratesby a nonspecialistrathersuperfluous.But
Tayles(1999) providesevidencefor highbirthratesamongfirstfarmersin Neolithic
centralThailand.Even withoutdirect skeletalevidence for fertilityincreases,the
cemetery size and site-areaincreases in the Neolithic/Formativein regions such
as the Levant,China, and Mesoamericatend to make it fairly obvious that there
were many more first farmersin agriculturalhomelandregions than there were
last hunters(Hassan 1981). In terms of diet, a numberof recent stable isotope
studies on bone have also pointed to majorchanges towardincreasingquantities
of agriculturalfoods across the transition(e.g., Richards& Hedges 1999, Bonsall
1997).
But perhapsthe main point to be derivedfrom the palaeoanthropologicalliter-
atureis that both populationcontinuityand replacementcan be identifiedacross
the hunter-to-farmer transitionin differentpartsof the world;there is no right or
wrong universal, and every situationneeds to be examinedon its own merits.
In terms of genetic history derived from living populations,we seem to be
currentlyin a phase of impasse. The recombiningnuclearDNA data for Europe,
as analyzed from a multisystemstandpoint,present a clinal situationsuggestive
of significantdemic diffusion across the continent from southeastto northwest
at some time in prehistory(Ammerman& Cavalli-Sforza1984, Cavalli-Sforza
et al 1994, Barbujaniet al 1994, 1998, Cavalli-Sforza& Minch 1997, Chikhi
et al 1998, Barbujani& Bertorelle 2001). These techniques in themselves of-
fer no time depth or precise ties with early farmersand are basically phenetic
ratherthan directly phylogenetic in implication,but they are neverthelesshighly
suggestive. As Cavalli-Sforza& Cavalli-Sforza(1995:149) point out, the first
principalcomponentof one particularEuropeangene frequencyanalysisexplains
28% of the total variationandmust reflectpopulationmovement-so many genes
 EARLYAGRICULTURAL
DIASPORAS? 199

could not produce such a dine as a result of naturalselection alone, unless we

follow the suggestion of Fix (1996, 1999) that the moving agropastoralsystem
itself was supplyingthe selective basis for the gene frequenciesvia zoonotic dis-
eases. Even from this viewpoint, however, Fix is unable to argue strongly for
hunter-gathereradoptionof agricultureas opposed to farmerspread,and indeed
Fix concludes that genetics alone cannot solve the problem.Furthermore,Krantz
(1988, p. 93) offers the interestingobservationthat a constant incorporationof
native hunter-gatherergenes into a spreadingagriculturalpopulationof ultimate
SouthwestAsian origin would give the eventualNorthwestEuropeanfirst farm-
ers a genome between 70% and 94% of Europeanderivation.So, even if demic
diffusion did occur across Europe during the Neolithic, we can hardly expect
the Neolithic British to have been exact clones of the denizens of Catalhoyukin
Anatolia.
The strongestclaims against the reconstructionsfavoringdemic spreadin the
EuropeanNeolithic come from studies of the nonrecombiningportions of the
genome-mitochondrial DNA andthe Y chromosome.In Europe,currentcalcula-
tions of coalescence times for mtDNA haplotypessuggest originationin European
Late Palaeolithicratherthan Neolithic Southwest Asian chronological contexts
(Sykes 1999, Richardset al 2000). But some very crucialissues here would seem
to involve the precise mutationrates utilized for such calculations. Indeed, my
impression of the nonrecombiningDNA literatureis that coalescence dates are
getting ever youngerand overlappingwith the terminalPleistocene andthe begin-
ning of the Holocene (e.g., Torroniet al 1998, Sykes 1999, Excoffier& Schneider
1999, especially Kayseret al 2001). I anticipatefurthermovementtowardan even
younger directionin the near future.There is tremendousscope here for insight
into earlyfarmingdispersal,particularlyif the impressivestarburstsof mtDNAlin-
eages presentedby Di Rienzo & Wilson (1991; see also Rogers & Jorde1995) can
be datedto the relevanttimescale. Kayseret al (2000, 2001) suggest majorphases
of populationexpansioncommencing6000 BP for Austronesiansand2200 BP for
Polynesians from Y chromosomedata, and it has to be admittedthat these dates
do fit extremelywell with the archaeologicaland linguistic data on Austronesian
dispersal.
In the Pacific region, recent work on the nonrecombininggenetic systems has
focusedattentionon Austronesianoriginsin IslandSoutheastAsia andTaiwan,ver-
sus Papuanorigins amongmore ancientpopulationsin westernOceania(Richards
et al 1998, Lum et al 1998, Merriwetheret al 1999, Hagelberget al 1999, Bing
Su et al 2000, Kayseret al 2000, 2001). This researchsupportsa generalscenario
of Austronesianhorticulturalistexpansionaroundthe Papuanhorticulturaland ar-
boriculturalhomelandregion of New Guinea, combined later with a substantial
Papuantake-overof Austronesianlanguage and genotype in the western Pacific
(Bellwood 1998, Kayseret al 2000, 2001). The latterbringsup a perhapsobvious
point, thata "native"genotypein an environmentwith a high incidenceof diseases
such as malariawill have some selective advantageover an immigrantgenotype.
200 BELLWOOD

But althoughthis perspectivecan explain why Melanesiais Melanesian,it cannot

explain the much strongerAsian genetic heritagein neighboringandequally trop-
ical westernandcentralIndonesia,a heritageclearlyvisible in all genetic systems.
Here we must fall back on culturalexplanations,one of these being thatmigrating
Austronesiansreplacedhunter-gatherers in Indonesiabutwere in turnabsorbedby
denserpopulationsof existing arboriculturalists in lowland Melanesia (Bellwood
1997a, 1998).
Becausethe geneticfieldof populationhistoryis in a constantstateof flux,as be-
fits its new andrevolutionaryexplosion into the arenaof prehistory,I avoidfurther
commentandmerelyraise the questionof how moder gene distributionsin living
populationscan reallyreflecton prehistoric"events"thatoccurredmanymillennia
ago. As Bertranpetit(2000, p. 6927) points out, in discussing an apparentmisfit
between Y chromosomeand palaeoanthropologicalchronologiesfor moder hu-
man expansion,inferencesfrommolecules to populationsarenot straightforward.
As Weiss (1998, p. 285) also pointsout, differentpopulationhistoriescan generate
the same genetic outcome.
If Polynesianor EuropeanmtDNAlineages indeedturnout to havePalaeolithic
coalescencetimes(as claimedby Richardset al 1998, 2000), we still needto askjust
wherethese lineages originallycommencedtheirexistences in geographicalterms
(Barbujani& Bertorelle2001). Therecouldbe problemsof maskedphylogenyhere
similarto those in archaeology,especially if the tracesof genetic events thatmight
have seemed majorat the time have become hidden by the vagariesof millennia
of subsequenthistory.

CONCLUSIONS:HOMELANDS,SPREADING
INTO FRICTION,AND BEYOND
In this paperI have obviously taken the position that Neolithic farmerdispersal
was an importantfactorin establishingthe currentworldpatternof languagesand
geographicalraces. I have also pointedto regions where such dispersalwas mini-
mized by hunter-gatherer adoptionof agricultureandlanguage(certainlyMelane-
sia, maybe westernand northernEurope,maybe northernIndia,but in all honesty
I find it hard to point with great conviction to many other large-scale regions).
Archaeologistshave greatdifficultyin coming to agreementon this issue, as can
be seen from the absolutely voluminous literaturethat has emanatedin recent
years on the Mesolithic to Neolithic transitionin Europe,favoringboth Neolithic
"packages"andMesolithicadoption.By contrast,Americanarchaeologyhas been
singularlyquiet on this issue, mainly because a majorityof North Americanar-
chaeologists acceptwithoutquestiona hunter-gatheradoptionof agriculturein all
situationsandrelativelyfew (with notableexceptions)takean interestin linguistic
prehistory.
Manylinguists,however,do supportagricultureandlanguagefarmer-dispersal
correlationsfor language family origins, to the extent that even the Trans-New
 EARLYAGRICULTURAL
DIASPORAS? 201

GuineaPhylumlanguagesof westernMelanesiaarenow being debatedas another

possible exampleof early farmerdispersal(PapuanPasts 2000; and see Hagelberg
et al 1999 for some mtDNA data that could supportthis). Some linguists remain
negative about farmerdispersal (Campbell 1999, p. 221), as do some biologists
(Oppenheimer1997; see Bellwood 2000b for a discussion of this book). But other
geneticists (Cavalli-Sforza& Cavalli-Sforza1995, Barbujaniet al 1998) seem to
have no doubtsaboutits efficacy as an explainerof the past.
In the future, increased knowledge and understandingwill only come from
careful multidisciplinaryconsiderationsof many strandsof evidence (Renfrew
1992, 2000). This observationapplies to archaeologists,linguists, palaeoanthro-
pologists and geneticists alike; disciplinary superioritygets us nowhere. Right
now, I sense a fairly even balance across the anthropologicalcommunity for
and against the concept of early farmingdispersals, a circumstancewhich sug-
gests to me that (a) there can be no absolute answer, and (b) reality combines
both perspectives of agriculturalistdispersal and hunter-gathereradoption.But
in any specific situation, reality will not be absolutely balanced, and the sci-
ences of prehistoryneed to treatevery situationas of equal significance.Starburst
zones, spreadzones, friction zones, and also the "beyond"zones where migrat-
ing former-farmerswere obliged to specialize into hunting and gathering,can
be expected to give entirely different results with respect to phylogenetic ver-
sus reticulativetransmissionof genes, languages,and materialcultures(Table 1).
Whatworksfor centralEuropemightnot workfor coastalNew Guineaor the Great
Basin.

TABLE1 Simplifiedcharacter
statesforthefourconceptualized
zonesof agricultural
origin
andspread
Homeland/Starburst Beyond
zones Spreadzones Frictionzones (noagriculture)

Tempoof spread Upwellingat Fast Slow Variable

variedrates
Extentof spread Upwellingwith Great Varies,generally Variable
radialspread limited
Suitabilityfor High High Low Nil
agriculture
Priorhunter- High(buthunters Low Oftenhigh, Variable
gatherer becomefarmers) especiallyalong
population coastlinesand
densities rivers
Mesolithic- Yes Unlikely, Yes,to varying No
Neolithic exceptat degrees
continuity entrypoint
202 BELLWOOD
ACKNOWLEDGMENTS
I would like to thank Colin Groves and Malcolm Ross for their comments on this
essay. All errors, alas, are surely mine.
Visit the Annual Reviews home page at www.AnnualReviews.org
LITERATURECITED
Adams WY, van GervenDP, Levy RS. 1978.
The retreatfrom migrationism.Annu. Rev.
Anthropol.7:483-32
AikhenvaldA. 1996. Areal diffusion in north-
west Amazonia-the case of Tariana.An-
thropol.Linguist.38:73-116
AmmermanAJ, Cavalli-SforzaLL. 1984. The
NeolithicTransitionand the Geneticsof Pop-
ulations in Europe.Princeton,NJ: Princeton
Univ. Press
Anderson AJ. 1989. Prodigious Birds. Cam-
bridge:CambridgeUniv. Press
Anthony D. 1990. Migration in archaeology:
the baby, and the bathwater.Am. Anthropol.
92:895-914
Anthony D. 1997. Prehistoricmigrationas so-
cial progress. In Migrations and Invasions
in ArchaeologicalExplanation,ed. J Chap-
man, H Hamerow,pp. 21-32. Oxford:BAR
Int. Seri. 664
BallardCL. 1981. Aspects of the linguistichis-
toryof SouthChina.AsianPerspect.24:163-
85
BarbujaniG, BertorelleG. 2001. Genetics and
the populationhistoryof Europe.Proc. Natl.
Acad. Sci. USA98:22-25
BarbujaniG, BertorelleG, ChikhiL. 1998. Ev-
idence for Paleolithic and Neolithic gene
flow in Europe.Am.J. Hum.Genet. 62:488-
91
BarbujaniG, PilastroA, de DomenicaS, Ren-
frew C. 1994. Genetic variation in North
Africa and Eurasia:Neolithic demic diffu-
sion vs. Paleolithiccolonization.Am.J. Phys.
Anthropol.95:137-54
Bar-Yosef 0. 1998. On the nature of tran- BellwoodP.2000a. Thetime depthof majorlan-
sitions. Cambridge Archaeol. J. 8:141-
63
Bellwood P. 1984. The greatPacific migration.
Encycl. Br. Yrbk.Sci. Futurefor 1984, pp.
80-93
Bellwood P. 1984-1985. A hypothesisfor Aus-
tronesian origins. Asian Perspect. 26:107-
17
Bellwood P. 1991. The Austronesiandisper-
sal and the origins of languages. Sci. Am.
265:88-93
Bellwood P. 1994. An archaeologist'sview of
language macrofamilyrelationships.Ocea-
nic Linguist.33:391-406
Bellwood P. 1996a. The origins and spreadof
agriculturein the Indo-Pacificregion.In The
Origins and Spreadof Agricultureand Pas-
toralisminEurasia,ed. D Harris,pp.465-98.
London:UCL Press
Bellwood P. 1996b.Phylogenyandreticulation
in prehistory.Antiquity70:881-90
Bellwood P. 1996c. Hierarchy,founder ideol-
ogy andAustronesianexpansion.In Origins,
Ancestryand Alliance, ed. J Fox, C Sather,
pp. 18-40. Canberra:Dep. Anthropol.,Res.
Sch. Pac. Asian Stud., Aust. Natl. Univ.
Bellwood P. 1997a. Prehistory of the Indo-
Malaysian Archipelago. Honolulu: Univ.
Hawaii Press
Bellwood P. 1997b. Prehistoricculturalexpla-
nations for widespread language families.
See McConvell& Evans, pp. 123-34
Bellwood P. 1998. From Bird's Head to bird's
eye view: long term structuresand trendsin
Indo-Pacific prehistory.In Perspectives on
the Bird's Head of Irian Jaya, Indonesia,
ed. J Miedema,C Od6, R Dam, pp. 951-75.
Amsterdam:Rodopi
guage families: an archaeologist'sperspec-
tive. In TimeDepth in HistoricalLinguistics,
ed. C Renfrew, A McMahon, L Trask, 1:
 EARLYAGRICULTURALDIASPORAS?
10940. Cambridge: McDonald Inst. Ar-
chaeol. Res.
Bellwood P. 2000b. Some thoughts on under-
standingthe human colonization of the Pa-
cific. People Cult. Oceania 16:5-17
Bellwood P. 2001. Archaeologyand the histor-
ical determinantsof punctuationin language
family origins. In Areal Diffusion and Ge-
netic Inheritance:Problemsin Comparative
Linguistics,ed. R Dixon, A Aikhenvald.Ox-
ford:OxfordUniv. Press. In press
Benz BL, LongA. 2000. Prehistoricmaize evo-
lution in the TehuacanValley.Curr.Anthro-
pol. 41:459-65
BertranpetitJ. 2000. Genome, diversity and
origins: the Y chromosome as a storyteller.
Proc. Natl. Acad. Sci. USA97:6927-29
Bing Su, UnderhillP, MartinsonJ, Saha N,
McGarveyS, et al. 2000. Polynesianorigins:
insightsfromthe Y chromosome.Proc. Natl.
Acad. Sci. USA97:8225-28
Blust R. 1991. Soundchangeandmigrationdis-
tance. In CurrentTrendsin Pacific Linguis-
tics, ed. R Blust, pp. 27-42. Canberra:Pac.
Linguist., Ser. C-117
Blust R. 1995a. The position of the Formosan
languages: method and theory in Austro-
nesian comparative linguistics. In Austro-
nesian StudiesRelating to Taiwan,ed. P Li,
D Ho, Y Huang,C Tsang, C Tseng, pp. 585-
650. Taipei: Acad. Sin., Inst. Hist. Philol.,
Symp. Ser. 3
Blust R. 1995b. The prehistoryof the Austro-
nesian-speakingpeoples: a view from lan-
guage. J. WorldPrehist. 9:453-510
Blust R. 1999. On the explanationof similar-
ity. Pap. pres. at NSF Conf., Entering New
Landscapes,Univ. FL, Gainesville, Feb.
Bogucki P. 1996. The spreadof early farming
in Europe.Am. Sci. 84:242-53
Bonsall C, LennonR, McSweeney K, Stewart
C, HarknessD, et al. 1997. Mesolithic and
early Neolithic in the IronGates:a palaeodi-
etary perspective. J. Eur. Archaeol. 5:50-
92
BrownP. 1998. The firstMongoloids?: another
look at Upper Cave 101, Liujiang and Mi-
natogawa1. Acta Anthropol.Sin. 17:260-75
203
Bulbeck D. 2000. Dental morphology at
Gua Cha, West Malaysia. Bull. Indo-Pac.
Prehist.Assoc. 19:17-41
Burmeister S. 2000. Archaeology and mi-
gration: approaches to an archaeological
proofof migration.Curr.Anthropol.41:539-
68
Campbell L. 1999. Nostratic and linguistic
palaeontology in methodological perspec-
tive. In Nostratic: Examining a Linguistic
Macrofamily,ed. C Renfrew, D Nettle, pp.
179-230. Cambridge:McDonald Inst. Ar-
chaeol. Res.
Cavalli-SforzaLL, Cavalli-SforzaF. 1995. The
Great Human Diasporas. Reading, MA:
Addison-Wesley
Cavalli-SforzaL, Menozzi P, Piazza L. 1994.
The History and Geography of Human
Genes. Princeton,NJ: PrincetonUniv. Press
Cavalli-SforzaL, Minch E. 1997. Paleolithic
and Neolithic lineages in the Europeanmi-
tochondrialgene pool. Am. J. Hum. Genet.
61:247-51
ChakrabartiD. 1999. India:An Archaeological
History. New Delhi: OxfordUniv. Press
ChapmanJ, Hamerow H, eds. 1997. Migra-
tions andInvasionsin ArchaeologicalExpla-
nation. Oxford:BAR Int. Ser. 664
ChikhiL, Destro-Bisol G, BertorelleG, Pas-
cali V, BarbujaniG. 1998. Clines of nuclear
DNA markerssuggest a largelyNeolithic an-
cestryof the Europeangene pool. Proc. Natl.
Acad. Sci. USA95:9053-58
Crosby AW. 1986. Ecological Imperialism.
Cambridge:CambridgeUniv. Press
Diamond J. 1997a. Guns, Germs and Steel.
London:JonathanCape
DiamondJ. 1997b. The language steamrollers.
Nature 389:544-46
Di Rienzo A, Wilson A. 1991 Branchingpat-
ternin the evolutionarytree for humanmito-
chondrialDNA. Proc. Natl. Acad. Sci. USA
88:1597-1601
Dixon RMW. 1997. The Rise and Fall of Lan-
guages. Cambridge:CambridgeUniv. Press
Dolgopolsky A. 1993. More about the Indo-
Europeanhomeland problem. Medit. Lang.
Rev. 6:230-48
204 BELLWOOD
Dutton T. 1995. Languagecontact and change
in Melanesia. In The Austronesians,ed. P
Bellwood,J Fox, D Tryon,pp. 192-213. Can-
berra:Dep. Anthropol.,Res. Sch. Pac. Asian
Stud., Aust. Natl. Univ.
EhretC. 1998. AnAfricanClassical Age. Char-
lottesville: Univ. Press Virginia
ErringtonJ. 1998. Shifting Languages. Cam-
bridge:CambridgeUniv. Press
ExcoffierL, SchneiderS. 1999.Why do hunter-
gathererpopulationsnot show signs of Pleis-
tocene demographicexpansion?Proc. Natl.
Acad. Sci. USA96:10597-602
Evans N, Jones R. 1997. The cradle of the
Pama-Nyungans.See McConvell & Evans,
pp. 385-417
Fix A. 1996. Gene frequencydines in Europe:
demic diffusion or natural selection? J. R.
Anthropol.Inst. 2:625-44
Fix A. 1999. Migration and Colonization in
Human Evolution. Cambridge:Cambridge
Univ. Press
Fortescue M. 1997. Dialect distributionand
small group interactionin GreenlandicEs-
kimo. See McConvell & Evans, pp. 111-
22
Fox CL. 1996.Physicalanthropologicalaspects
of the Mesolithic-Neolithictransitionin the
IberianPeninsula.Curr.Anthropol.37:689-
94
Gamkrelidze TV, Ivanov VV. 1995. Indo-
European and the Indo-Europeans.Berlin:
Moutonde Gruyter
Gronenbor D. 1999. A variation on a basic
theme: the transitionto farmingin southern
centralEurope.J. WorldPrehist. 13:123-210
HagelbergE, KayserM, Nagy M, RoewerL,
Zimdahl H, et al. 1999. Molecular genetic
evidence for the humansettlementof the Pa-
cific. Philos. Trans.R. Soc. London Ser. B
354:141-52
Hall M. 1990. Farmers, Kings and Traders.
Chicago:Univ. Chicago Press
HarrisD. ed. 1996. The Origins and Spreadof
AgricultureandPastoralismin Eurasia.Lon-
don: UCL Press
HarrisM. 1978. Cannibalsand Kings. London:
Collins
Hassan F. 1981. Demographic Archaeology.
New York:Academic
HeathS, LapradeR. 1982. Castiliancoloniza-
tion and indigenous languages:the cases of
Quechuaand Aymara.In Language Spread,
ed. RL Cooper, pp. 118-47. Bloomington:
IndianaUniv. Press
Higham C. 1996. Archaeology and linguistics
in SoutheastAsia: implicationsof the Austric
hypothesis. Bull. Indo-Pac. Prehist. Assoc.
14:110-18
Hill J. 1999. Whyis Uto-Aztecanso big? Pre-
sentedat Dep. Anthropol.Colloq. Ser.,Univ.
Calif. Davis
HudsonM. 1999. Ruins of Identity.Honolulu:
Univ. Hawai'i Press
Hunn ES, Williams NM. 1982. Introduction.
In ResourceManagers:NorthAmericanand
AustralianHunter-Gatherers,ed. NM Wil-
liams, ES Hunn, pp. 1-16. Boulder, CO:
Westview
JackesM, LubellD, MeiklejohnC. 1997a. On
physical anthropological aspects of the
Mesolithic-Neolithic transition in the Ibe-
rianPeninsula.Curr.Anthropol.38:839-46
Kayser M, BrauerS, Weiss G, Underhill P,
Roewer L, et al. 2000. Melanesian origin
of Polynesian Y chromosomes. Curr.Biol.
10:1237-46
KayserM, BrauerS, Weiss G, Schiefenhovel
W, UnderhillA, StonekingM. 2001. Inde-
pendenthistoriesof humanY chromosomes
from Melanesia and Australia.Am. J. Hum.
Genet. 68:173-90
Keeley LH. 1992. The introductionof agricul-
tureto the westernNorthEuropeanPlain. In
Transitionsto Agriculturein Prehistory,ed.
AB Gebauer,D Price, pp. 81-96. Madison:
PrehistoryPress
Kent S ed. 1989. Farmers as Hunters. Cam-
bridge:CambridgeUniv. Press
Kent S. 1992. The currentforagercontroversy.
Man 27:45-70
Kirch PV. 1997. The Lapita Peoples. Oxford:
Blackwell
KirchPV, GreenRC. 2001. Hawaiki:Ancestral
Polynesia.An Essay in HistoricalAnthropol-
ogy. Cambridge:CambridgeUniv. Press
 EARLYAGRICULTURALDIASPORAS?
Krantz G. 1988. The Geographical Develop-
mentof EuropeanLanguages.New York:Pe-
ter Lang
Kulick D. 1992. Language Shift and Cultural
Reproduction.Cambridge:CambridgeUniv.
Press
LarsenCS. 1995. Biological changes in human
populationswith agriculture.Annu.Rev.An-
thropol.24:185-213
Lum JK, CannR, MartinsonJ, JordeL. 1998.
Mitochondrialand nuclear genetic relation-
ships among Pacific Island and Asian popu-
lations.Am.J. Hum. Genet. 63:613-24
Mallory J. 1989. In Search of the Indo-
Europeans.London:Thames& Hudson
Mallory J. 1997. The homelands of the Indo-
Europeans.In Archaeologyand LanguageI,
ed. R Blench, M Spriggs, pp. 93-121. Lon-
don: Routledge
Masica C. 1978. Aryan and Non-Aryan ele-
ments in North Indianagriculture.In Aryan
andNon-Aryanin India, ed. M Deshpande,P
Hook, pp 55-152. Ann Arbor:Univ. Mich.,
Pap. South and SoutheastAsia 14
McConvellP. 1996. Backtrackingto Babel.Ar-
chaeol. Oceania 31:125-44
McConvellP, EvansN, eds. 1997.Archaeology
and Linguistics. Melbourne,Aust.: Oxford
Univ. Press
Meacham W. 1984-1985. On the improbabil-
ity of Austronesianorigins in South China.
Asian Perspect.26:89-106
Meindl R, Russell K. 1998. Recent advances
in method and theory in paleodemography.
Annu.Rev.Anthropol.27:375-99
MerriwetherD, FriedlaenderJ, MediavillaJ,
Mgone C, Gentz F, FerrellR. 1999. Mito-
chondrialDNA variationis an indicatorof
Austronesianinfluence in Island Melanesia.
Am.J. Phys. Anthropol.110:243-70
MiihlhiiuslerP. 1986. Pidgin and Creole Lin-
guistics. Oxford:Blackwell
MurdockGP. 1967. EthnographicAtlas. New
Haven, CT:HRAF Press
MuroM. 1998 (1989). Notjust anotherroadside
attraction.Am.Archaeol. 2/4:10-16
Nettle D. 1996. Language diversity in West
Africa.J. Anthropol.Archaeol. 15:403-38
205
Nettle D. 1999. Linguistic Diversity. Oxford:
OxfordUniv. Press
Nichols J. 1992. Language Diversity in Space
and Time.Chicago:Univ. Chicago Press
Nichols J. 1997. Modeling ancient population
structures and movements in linguistics.
Annu.Rev.Anthropol.26:359-84
Nichols J. 1998. The Eurasianspreadzone and
the Indo-Europeandispersal.InArchaeology
and Language II, ed. R Blench, M Spriggs,
pp. 220-66. London:Routledge
OppenheimerS. 1997. Eden in the East. Lon-
don: Weidenfeld& Nicholson
Papuan Pasts 2000. Conference on the cul-
tural, linguisticand biological historyof the
Papuan speakingpeoples, Nov. 27-30, Res.
Sch. Pac. Asian Stud.,Aust. Natl. Univ.
Pawley AK. 1999. Chasing rainbows: impli-
cations of the rapid dispersal of Austrone-
sian languages. In Selected Papersfrom the
Eighth International Conference on Aus-
tronesianLinguistics,ed. E Zeitoun,P Li, pp.
95-138. Taipei:Inst. Linguist.,Acad. Sin.
Pawley AK, Ross M. 1993. Austronesianhis-
torical linguistics and culturehistory.Annu.
Rev.Anthropol.22:425-59
Pawley AK, Ross M, eds. 1994. Austronesian
Terminologies,Continuityand Change.Can-
berra:Pac. Linguist. Ser. C-127
Pawley AK, Ross M. 1995. The prehistoryof
theOceaniclanguages.InTheAustronesians,
ed. P Bellwood, J Fox, D Tryon,pp. 39-74.
Canberra:Dep. Anthropol., Res. Sch. Pac.
Asian Stud., Aust. Natl. Univ.
Peiros I. 1998. Comparative Linguistics in
SoutheastAsia.Canberra:Pac. Linguist.Ser.
C-142
Pentz P. 1992. The InvisibleConquest.Copen-
hagen:Natl. Mus. Denmark
PetersonJ. 1978. TheEcology of Social Bound-
aries. Urbana:Univ. Ill. Press
Phillipson D. 1985. An archaeologicalrecon-
siderationof Bantuexpansion.Muntu2:69-
84
PhillipsonD. 1993.AfricanArchaeology.Cam-
bridge:CambridgeUniv. Press. 2nd ed.
Price TD. 1995. Social inequality at the ori-
gins of agriculture.In Foundationsof Social
206 BELLWOOD
Inequality, ed. TD Price, GF Feinman, pp.
129-51. New Yok:Plenum
Price TD, ed. 2000. Europe's First Farmers.
Cambridge:CambridgeUniv. Press
PriceTD, GebauerAB. 1992. Foragersto farm-
ers: an introduction.In Transitionsto Agri-
culture in Prehistory, ed. AB Gebauer, D
Price, pp. 1-10. Madison:PrehistoryPress
Price TD, GebauerAB, eds. 1995. Last Hun-
ters, First Farmers.SantaFe, NM: Sch. Am.
Res.
RenfrewC. 1987. Archaeologyand Language.
London:JonathanCape
Renfrew C. 1991. Before Babel: speculations
on the origins of linguistic diversity.Cam-
bridgeArchaeol.J. 1:3-23
Renfew C. 1992. Worldlanguages and human
dispersals:a minimalistview. In Transitionto
Modernity,ed. JAHall, IC Jarvie,pp. 11-68.
Cambridge:CambridgeUniv. Press
Renfrew C. 1994. World linguistic diversity.
Sci. Am. 270:747-60
Renfrew C. 1996. Language families and the
spreadof farming.In TheOriginsand Spread
of Agricultureand Pastoralism in Eurasia,
ed. D Harris,pp. 70-92. London:Univ. Coll.
LondonPress
RenfrewC. 1999. Timedepth,convergencethe-
ory, and innovationin Proto-Indo-European.
J. Indo-Eur.Stud.27:257-93
Renfrew C. 2000. At the edge of knowability:
towards a prehistory of languages. Cam-
bridgeArchaeol.J. 10:7-34
RenfrewC, McMahonA, TraskL, eds. 2000.
TimeDepth in Historical Linguistics. Cam-
bridge: McDonald Inst. Archaeol. Res. 2
Vols.
Renfrew C, Nettle D, eds. 1999. Nostratic:
Examininga LinguisticMacrofamily.Cam-
bridge:McDonaldInst. Archaeol.Res.
RichardsM, Hedges R. 1999. A Neolithic rev-
olution?New evidence of diet in the British
Neolithic. Antiquity73:891-96
RichardsM, MacaulayV, Hickey E, Vega E,
Sykes B, et al. 2000. Tracing European
founderlineages in the NearEasternmtDNA
pool. Am.J. Hum. Genet. 67:1251-76
RichardsM, OppenheimerS, Sykes B. 1998.
MtDNA suggests Polynesianorigins in east-
ern Indonesia.Am.J. Hum.Genet. 63:1234-
36
Rogers A, Jorde L. 1995. Genetic evidence
on moder humanorigins.Hum.Biol. 67:1-
36
Rollefson GO, Kohler-Rollefson I. 1993.
PPNC adaptationsin the firsthalf of the 6th
millenniumBC. Pale'orient19:33-42
RosenbergM. 1994. Pattern,process, andhier-
archy in the evolution of culture.J. Anthro-
pol. Archaeol. 12:75-119
Ross M. 1988. Proto Oceanic and the Austro-
nesian Languages of Western Melanesia.
Canberra:Pac. Linguist.Ser. C-98
Ross M. 1991. How conservativeare sedentary
languages?In CurrentTrendsin PacificLin-
guistics, ed. R Blust, pp. 433-57. Canberra:
Pac. Linguist. Ser. C-117
Ross M. 1996. Contact-inducedchange and
the comparativemethod: cases from Papua
New Guinea. In The ComparativeMethod
Reviewed, ed. M Durie, M Ross, pp. 180-
217. New York:OxfordUniv. Press
Ross M. 1997a. Social networks and kinds
of speech communityevent. In Archaeology
and Language I, ed. R Blench, M Spriggs,
pp. 209-62. London:Routledge
Ross M. 1997b. Comment. Curr. Anthropol.
38:182-84
Ross M, PawleyA, OsmondM, eds. 1998. The
Lexicon of Proto Oceanic. 1. Material Cul-
ture.Canberra:Pac.Linguist.Ser.C, Vol. 152
Rouse I. 1986. Migrations in Prehistory.New
Haven, CT:Yale Univ. Press
Sapir E. 1918. Time Perspective in Aborigi-
nal AmericanCulture.Canada,Dep. Mines,
Geol. Surv.Mem. 90
Sather C. 1995. Sea nomads and rainforest
hunter-gatherers. In TheAustronesians,ed. P
Bellwood, J Fox, D Tryon,pp. 229-68. Can-
berra:Dep. Anthropol.,Res. Sch. Pac. Asian
Stud., Aust. Natl. Univ.
Sauder G. 1990. Europe Between the Lan-
guages. Canberra:Aust. Acad. Humanit.
Schooling S. 1990. Language Maintenancein
New Caledonia. Dallas, TX: Summer Inst.
Linguist.
 EARLYAGRICULTURALDIASPORAS?
SherrattA. 1997. Climatic cycles and behav-
ioral revolutions.Antiquity71:271-87
SherrattA, SherrattE. 1988. The archaeology
of Indo-European:an alternativeview.Antiq-
uity 62:584-95
Smalley WA. 1994. Linguistic Diversity and
National Unity. Chicago: Univ. Chicago
Press
Smith B. 1995. The Emergenceof Agriculture.
New York:Sci. Am.
Smith KP. 1995. Landnarm:the settlement of
Iceland in archaeologicaland historicalper-
spective. WorldArchaeol.26:319-47
Sorenson AP. 1982. Multilingualism in the
northwestAmazon.In Sociolinguistics,ed. J
Pride,J Holmes,pp. 78-93. Harmondsworth:
Penguin
Spriggs M. 1996. What is Southeast Asian
aboutLapita?In PrehistoricMongoloidDis-
persals, ed. T Akazawa, E Szathmary,pp.
324-48. Oxford:OxfordUniv. Press
SteadmanD. 1999. The prehistoricextinction
of South Pacific birds. In The Pacific from
5000 to 2000 BP, ed. J-C Galipaud,I Lilley,
pp. 375-86. Paris:Inst. Rech. Dev.
SverdrupH, GuardansR. 1999. Compiling
words from extinct Non-Indoeuropeanlan-
guages in Europe. In Historical Linguistics
and Lexicostatistics,ed. V Shevoroshkin,P.
Sidwell, pp. 201-57. Melbourne:AHL
Sykes B. 1999. The moleculargenetics of Euro-
pean ancestry.Philos. Trans.R. Soc. London
Ser.B354:131-40
Tayles N. 1999. The Excavation of Khok
Phanom Di. Vol. V: The People. London:
Soc. Antiquaries
Thomason SG, KaufmanT. 1988. Language
Contact,Creolization,and Genetic Linguis-
tics. Berkeley:Univ. Calif. Press
TorroniA, BandeltH-G, D'UrbanoL, Laher-
mo P, MoralP, et al. 1998. MtDNA analysis
revealsa majorlatePaleolithicpopulationex-
207
pansion from southwesternto northeastern
Europe.Am.J. Hum. Genet. 62:1137-52
VanAndel T, ZanggerE, DemitrackA. 1990.
Land use and soil erosion in prehistoricand
historic Greece. J. Field Archaeol. 17:379-
96
Van Andel T, Runnels C. 1995. The earliest
farmersin Europe.Antiquity69:481-500
VansinaJ. 1990. Paths in the Rainforest.Madi-
son: Univ. Wisc. Press
Vencl S. 1988. The role of hunting-gathering
populations in the transitionto farming: a
centralEuropeanperspective.In Hunters in
Transition,ed. M Zvelebil, pp. 43-51. Cam-
bridge:CambridgeUniv. Press
Weiss K. 1998. Coming to terms with human
variation. Annu. Rev. Anthropol. 27:273-
300
Wiik K. 2000. Some ancient and moder lin-
guisticprocessesin northernEurope.In Time
Depth in Historical Linguistics, ed. C Ren-
frew,A McMahon,L Trask,2:463-80. Cam-
bridge:McDonaldInst. Archaeol.Res.
Wood J, Milner G, HarpendingH, Weiss K.
1992. The osteological paradox. Curr.An-
thropol.33:343-70
Woodbur J. 1982. Egalitariansocieties. Man
17:431-51
Zvelebil M. 1995. At the interfaceof archaeol-
ogy, linguistics and genetics:Indo-European
dispersals and the agriculturaltransitionin
Europe.J. Eur.Archaeol.3:33-70
Zvelebil M. 1998. Agriculturalfrontiers,Ne-
olithic origins, and the transition to farm-
ing in the Baltic region. In Harvesting the
Sea, Farming the Forest, ed. M Zvelebil, L
Domanska, R Dennell, pp. 9-27. Sheffield:
SheffieldAcad. Press
Zvelebil M, Rowley-ConwyP. 1986. Foragers
andfarmersin AtlanticEurope.InHuntersin
Transition,ed. M Zvelebil, pp. 67-95. Cam-
bridge:CambridgeUniv. Press